- Published
- Updated
A change in a biomarker or intermediate measurement is not identical to an outcome describing how participants feel, function or survive. Selecting a surrogate requires an explicit statement of what the finding can and cannot establish.
Define the question and assumptions
Separate biological rationale, analytical reliability and evidence connecting the marker to the intended clinical outcome. An observed association does not necessarily show that changing the marker through an intervention changes that outcome. Consider the disease, population and intervention context in which supporting evidence was obtained.
Hypothetical study example
In a hypothetical early-phase study, a laboratory marker improves but functional follow-up is insufficient. The team reports the marker change as observed. Rather than claiming an unmeasured long-term benefit, it identifies the clinical questions that a subsequent study would need to address directly.
Design review checks
- Give the marker and intended clinical outcome separate names and definitions.
- Examine whether validation evidence transfers to the proposed research context.
- State which clinical interpretations are supported by the available follow-up duration.
Record the decision and its limits
The accepted rationale explains both measurement selection and unsupported inferences. Avoid promises of regulatory acceptance or approval. Do not hide discordance between outcomes: consider biological expectations, measurement limitations and clinical effects separately, and carry the remaining uncertainty through to the report and its conclusions.
Frequently asked questions
- Is biological plausibility sufficient by itself?
- No. Measurement quality and the connection to clinical interpretation need separate assessment.
- Can surrogate and clinical outcomes share a table?
- Yes, provided their meanings and inferential limits are clearly distinguished.